Abstract Background The addition of relevant parameters to acute kidney injury (AKI) criteria might allow better prediction of patient mortality than AKI criteria alone. Here, we evaluated whether inclusion of AKI duration could address this issue. Methods AKI was defined according to the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines in 2,143 critically ill patients, within 15 days of patient admission. AKI cases were categorized according to tertiles of AKI duration: 1^st tertile, 1–2 days; 2^nd tertile, 3–5 days; and 3^rd tertile, ≥6 days. The hazard ratios (HRs) for overall survival rates in three groups were calculated after adjustment for multiple covariates compared with ICU patients without AKI as the reference group. The predictive ability for mortality was assessed by calculating the area under the curve (AUC) of the receiver operating characteristic curve. Results AKI increased the HRs for overall mortality, and the mortality rate increased with AKI duration: the adjusted HRs were 1.99 (1^st tertile), 2.67 (2^nd tertile), and 2.85 (3^rd tertile) compared with the non-AKI group (all Ps < 0.001). The AUC of the ROC curve for overall mortality based on the AKI duration groups (0.716) was higher than the AUC of AKI staging using the KDIGO guidelines (0.696) (P = 0.001). When considering KDIGO stage and AKI duration together, the AUC (0.717) was also significantly higher than that using the KDIGO stage alone (P < 0.001). Conclusions AKI duration is an additional parameter for the prediction of mortality in critically ill patients. The inclusion of AKI duration could be considered as a refinement of the AKI criteria.